Phosphoacceptor Site S173 in the Regulatory Domain of Epstein-Barr Virus ZEBRA Protein Is Required for Lytic DNA Replication but Not for Activation of Viral Early Genes

Expression vectors.The ZEBRA (pHD1013/Z) and Rta (pRts/Rta) expression vectors were previously described (7). The QuikChange site-directed mutagenesis system (Stratagene) was used to mutate ZEBRA at positions 167 and 173. The sequences of mutagenic primers are available upon request. BALF2 was amplified from EBV B95-8 DNA, using a primer set containing an EcoRI site and an XbaI site at the 5′ and 3′ ends, respectively (CTTGAATTCCATGCAGGGTGCACAGACTAGCGAGG and CAGTCTAGACTAGACCTCGAGTCCGGGGAGAACGG). The amplified genomic fragment was cloned into pFLAG-CMV2 (a kind gift from Ren Sun).

Cell lines.BZKO (Bam Z knockout) is a 293 human embryonic kidney cell line stably transfected with an EBV bacmid in which the BZLF1 gene was disrupted by inserting the hygromycin resistance gene (11). HH514-16 is an EBV-positive cell line derived from a human Burkitt's lymphoma (19). HKB5/B5 is an EBV-negative subclone that was originally generated by fusing HH514-16 cells with 293 human embryonic kidney cells (4).

Activation of the EBV lytic cycle.Activation of the lytic cycle in BZKO cells was attained by ectopic expression of the BZLF1 gene. Transfection conditions for BZKO cells have been described (18). HH514-16 cells in logarithmic growth phase, usually 48 h after subculture, were induced to enter the EBV lytic cycle by adding 3 mM sodium butyrate (Sigma). Cells were assessed for the expression of mRNAs of EBV replication genes 48 h after chemical treatment.

Protein extracts and Western blot analysis.Harvested cells were resuspended in sodium dodecyl sulfate (SDS) sample buffer at 10⁶ cells/10 μl. Proteins were separated in 10% SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Bio-Rad). Rabbit polyclonal antibodies were used to detect ZEBRA (46), Rta (35), and BLRF2 (21). EBNA1 and BFRF3 were detected by the human antiserum SJ (41). EA-D was detected by the mouse monoclonal antibody R3 (5). Antibody-protein complexes were detected using ¹²⁵I-labeled protein A.

Northern blot analysis.Cellular RNAs were prepared from 10⁷ cells by using an RNeasy purification kit (QIAGEN). ³²P-labeled probes were prepared using random hexanucleotides to prime DNA synthesis. Full-length BALF2, BSLF1, and BBLF2/3 served as templates for random priming. BALF2 was excised from the expression vector BALF2/pFLAG-CMV2 by using EcoRI and XbaI. Full-length BALF5, BSLF1, and BBLF2/3 were excised from their corresponding expression vectors (pRTS13, pRTS11, and pRTS25, respectively) by using XbaI and HindIII. The XhoI restriction fragment, containing 513 bp (positions 914 to 1456) of BBLF4, was excised from pRTS28. The expression vectors for BALF5, BBLF2/3, BBLF4, and BSLF1 were a kind gift from Diane Hayward. A radiolabeled 370-bp NcoI-PstI fragment of the H1 component of RNase P was used as a probe to control for RNA loading.

EMSA.Cell extracts for electrophoretic mobility shift assays (EMSAs) were prepared as described previously (6). Annealed oligonucleotides, each containing one of the seven ZEBRA response elements present in oriLyt (27) or a ZIIIB site (6), were end labeled with γ-³²P, using T4 polynucleotide kinase (NEB). DNA binding reactions were prepared and analyzed as previously described (6). Supershifts were performed using BZ1, a murine monoclonal antibody against a region in the C-terminal domain of ZEBRA that spans amino acids 214 to 230 (18, 48).

EBV lytic replication assay.BZKO cells transfected with expression vectors encoding wild-type or mutant ZEBRA proteins were harvested 48 h after transfection. Total DNA was extracted as previously described (42). Ten micrograms of DNA was digested with 40 units of BamHI at 37°C for 3 h. The DNA fragments were separated in a 0.8% agarose gel in 1× Tris-borate-EDTA and transferred to a Zeta probe GT genomic membrane (Bio-Rad) by using 0.4 N NaOH. The separated BamHI restriction fragments encompassing the viral terminal repeats were detected using a 1.9-kb XhoI probe complementary to a unique region upstream of the repeats.

ChIP assay.DNA-protein complexes formed in ZEBRA-transfected BZKO cells (∼5 × 10⁶ cells) were cross-linked by adding 1% formaldehyde to the growth medium. The cells were incubated for 10 min at 37°C and then washed once in phosphate-buffered saline containing protease inhibitors (Roche). Cells were resuspended in SDS lysis buffer (50 mM Tris-HCl [pH 8.1], 1% SDS, and 10 mM EDTA) and sonicated four times for 10 s each, using a Sonifier 450 apparatus (Branson). Cell lysates were cleared by centrifugation, and the collected supernatants were diluted 10-fold in chromatin immunoprecipitation (ChIP) dilution buffer (16.7 mM Tris-HCl [pH 8.1], 0.01% SDS, 1.1% Triton X-100, 167 mM NaCl, and 1.2 mM EDTA). A polyclonal rabbit antibody against ZEBRA was used to immunoprecipitate the ZEBRA-DNA complexes. The immune complexes were collected on protein G agarose (Upstate). The immunoprecipitated DNAs were amplified and quantitated by real-time PCR, using a Smart Cycler II machine (Cepheid). The relative concentration of DNA was calculated based on a standard curve constructed from different concentrations of oriLyt. To control for the amount of viral DNA present in each sample, the relative concentration of oriLyt immunoprecipitated by the anti-ZEBRA antibody was divided by the amount of oriLyt present in each input. A similar approach was followed to analyze ZEBRA binding to the BRLF1 promoter (Rp).

Solubility and DNase I treatment.To examine the solubility of ZEBRA, 2 × 10⁶ BZKO cells were resuspended in EMSA lysis buffer containing 20 mM HEPES (pH 7.5), 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM dithiothreitol, 25% glycerol, and protease inhibitors (Roche) in addition to X M NaCl, where X was equal to 0.3, 0.42, 0.6, or 1. Supernatants were separated from pellets by centrifugation at 90,000 rpm in a TLA 100 rotor in a benchtop TLX ultracentrifuge for 15 min at 4°C. Protein concentrations were determined using the Bradford reagent (Bio-Rad). Fifty micrograms of total protein was separated in an SDS-polyacrylamide gel and transferred to a nitrocellulose membrane. The amount of ZEBRA extracted was assessed by incubating the membrane with the polyclonal antibody against ZEBRA.

To determine the effects of DNase I treatment on the solubilities of different ZEBRA mutants, cell lysates prepared in EMSA lysis buffer containing 0.3 M NaCl were diluted fourfold in 20 mM HEPES (pH 7.5) and 2 mM MgCl₂. Extracts were incubated with 100 units of DNase I (Roche) at 37°C for 30 min. The solubilized form of ZEBRA was separated by ultracentrifugation, and the abundance of the soluble form was assessed by Western blot analysis.

ZEBRA CK2 site mutant Z(S167A/S173A) activates EBV early gene expression but not EBV late genes.To understand the role of phosphorylation of ZEBRA in the temporal regulation of the entire viral lytic cycle, we expressed wild-type (wt) ZEBRA and the Z(S167A/S173A) mutant in BZKO cells that carry an EBV genome in which the BZLF1 gene, encoding ZEBRA, has been inactivated by insertional mutagenesis (11). Thus, downstream lytic EBV gene expression in BZKO cells results solely from the introduction of a ZEBRA expression vector. The CK2 site mutant Z(S167A/S173A) was fully competent to activate the expression of Rta, the other early viral lytic cycle transcription factor, and EA-D, the EBV-encoded DNA polymerase processivity factor (Fig. 1). The Rta gene is a target of the direct action of ZEBRA, while EA-D is activated by the synergistic action of ZEBRA plus Rta. Therefore, CK2 site mutations do not affect the capacity of ZEBRA to activate viral target genes that are expressed early in the viral lytic cycle. Remarkably, however, the Z(S167A/S173A) mutant was deficient in the ability to activate expression of the BFRF3 gene, a late gene encoding a viral capsid component (Fig. 1). The Z(S167A/S173A) mutant was also defective at activating expression of another late gene, the BLFR2 gene, which encodes a tegument component (20; data not shown). Overexpression of Rta failed to restore the defect in this mutant, namely, a lack of activation of late genes (data not shown). These findings prompted us to examine the effect of abolishing the CK2 sites in lytic viral DNA replication.

• Open in new tab
• Download powerpoint

FIG. 1.

The CK2 site mutant Z(S167A/S173A) is defective at activating EBV late gene expression. BZKO cells were transfected with expression vectors for ZEBRA or the Z(S167A/S173A) mutant. Cell extracts were prepared at daily intervals, and immunoblots were probed for the indicated EBV gene products, i.e., EBNA1, Rta, EA-D, ZEBRA, and FR3.

Alanine substitution at S173 abrogates lytic viral DNA replication.EBV late gene expression is contingent on EBV lytic DNA replication (2, 42). Therefore, we determined whether mutation of the CK2 sites affected the capacity of ZEBRA to activate viral lytic DNA replication. EBV replicates in the form of a concatemer in which viral genomes are fused head to tail. Each viral genome in the concatemer has a variable number of ∼530-bp repeats at each terminus. During end processing, the synthesized concatemer is randomly cleaved between any two terminal repeats to produce single forms of the viral genome that vary in the number of terminal repeats. Therefore, the appearance of a ladder of DNA restriction fragments resulting from heterogeneous processing of the ends of the viral genome is a classical indicator of EBV lytic DNA replication (34) (Fig. 2A). In this assay, the Z(S173A) mutant, in which the principal CK2 site in ZEBRA was destroyed, was markedly impaired in the capacity to activate lytic viral DNA replication. The single Z(S167A) mutant, which has an altered minor CK2 site, was minimally affected. The phenotype of the double mutant Z(S167A/S173A) was similar to that of the Z(S173A) mutant alone (Fig. 2A). It can be seen from Fig. 2A that the same amount of input plasmid was detected in all lanes and that the amounts of ZEBRA expressed were equivalent (Fig. 2B).

• Open in new tab
• Download powerpoint

FIG. 2.

The CK2 site mutants Z(S173A) and Z(S167A/S173A) are defective at activating EBV lytic DNA replication. (A) BZKO cells were transfected with the indicated expression vectors. DNAs harvested after 48 h were analyzed for EBV DNA replication by Southern analysis with an “Xho 1.9” probe. *, input plasmids. (B) Cell extracts from the same transfections were analyzed for expression of EBNA1, Rta, and ZEBRA by immunoblotting. (C and D) BZKO cells were transfected with the indicated ZEBRA expression vectors in the absence or presence of PAA. DNAs harvested after 48 h were analyzed for EBV DNA content by quantitative PCR for oriLyt (C) or Rp (D).

The extent of amplification of EBV viral DNA following transfection of wt ZEBRA and the Z(S167A/S173A) mutant into BZKO cells was quantitated by real-time PCR (Fig. 2C and D), using primers for EBV oriLyt (Fig. 2C) and the EBV BRLF1 promoter Rp (Fig. 2D). A ZEBRA mutant, Z(S186E), that is deficient in binding to DNA and does not activate the EBV lytic cascade was studied in parallel (6). DNA binding of ZEBRA to the lytic origin of DNA replication, oriLyt, is required to activate EBV lytic DNA replication (10). In comparison to wt ZEBRA, the Z(S167A/S173A) mutant was 10.3- to 11.9-fold impaired in its capacity to amplify EBV DNA.

Alanine substitution at S173 does not impair expression of genes required for viral replication.While ZEBRA directly activated the expression of one early gene encoding a component of the replication machinery, namely, the BMRF1 gene (Fig. 1), it was conceivable that alanine substitution at S173 reduced the capacity of ZEBRA to activate the expression of other replication proteins. To investigate this possibility, we compared the mRNA levels of BALF2, BBLF2/3, BBLF4, BSLF1, and BALF5 expressed in BZKO cells transfected with wt ZEBRA, the Z(S173A) mutant, and the phosphomimetic substitution mutant Z(S173D). All three ZEBRA proteins were equally competent to activate the expression of these replication genes (Fig. 3 and data not shown).

• Open in new tab
• Download powerpoint

FIG. 3.

The Z(S173A) mutant is competent at activating the expression of early genes encoding the viral replication machinery. Northern blot analysis shows the levels of mRNAs transcribed from viral genes encoding the single-stranded DNA binding protein (BALF2), the primase-associated factor (BBLF2/3), the primase (BSLF1), and the helicase (BBLF4). Lanes 1, 2, and 3 represent HH514-16 cells that were untreated, (Unind.) treated with sodium butyrate (Ind.), and treated with sodium butyrate and PAA, respectively. Lanes 4, 5, 6, and 7 represent BZKO cells transfected with empty vector (cytomegalovirus) or an expression vector encoding wt ZEBRA, the Z(S173A) mutant, or the Z(S173D) mutant, respectively.

The Z(S167A/S173A) mutant is impaired in binding to EBV DNA in vitro.The remaining experiments attempted to account for the failure of the Z(S167A/S173A) mutant to stimulate DNA replication. The first approach was to compare in vitro binding of the wild type and the Z(S167A/S173A) mutant expressed in mammalian cells to each of the seven ZEBRA response elements present in oriLyt. Binding of the Z(S167A/S173A) mutant to ZRE-2, -4, and -5 was slightly impaired compared to that of the wild type (Fig. 4). However, wt ZEBRA and the Z(S167A/S173A) mutant appeared to bind equally to ZRE-1, -3, -6, and -7. Wild-type ZEBRA had a marginally higher affinity for ZIIIB, a ZEBRA response element located in the promoter of the BZLF1 gene, the gene encoding the ZEBRA protein, than that of the Z(S167A/S173A) mutant.

• Open in new tab
• Download powerpoint

FIG. 4.

Comparison of DNA binding of ZEBRA and the Z(S167A/S173A) mutant to the ZEBRA response elements that are present in oriLyt. Shown are the shifted portions of electrophoretic mobility shift assays using extracts of BZKO cells as a source of proteins and the seven different ZREs present in oriLyt as probes (27). The ZRE ZIIIB oligonucleotide was used as a positive control. A monoclonal antibody (BZ1) was added in alternate lanes to supershift the DNA-protein complexes. CMV, cytomegalovirus.

In the course of these experiments, it became apparent that the amount of immunoreactive Z(S167A/S173A) protein that was solubilized in 0.42 M NaCl, the conditions for making the cell extracts used for in vitro DNA binding assays, was considerably higher (fourfold in the experiment illustrated in Fig. 5A) than that of wt ZEBRA. Based on a dilution series, the CK2 site mutant Z(S167A/S173A) was impaired approximately fourfold in binding to a ZRE-2 site (Fig. 5B). Correction for the amount of immunoreactive ZEBRA protein present in the Z(S167A/S173A) cell extract revealed that the mutant was markedly impaired at binding DNA in vitro. When cell extracts were diluted 1:4 in extracts of cells transfected with empty vector in order to equalize the amount of immunoreactive ZEBRA protein, the Z(S167A/S173A) mutant could no longer be shown to bind to a ZRE-2 site (Fig. 5C and D).

• Open in new tab
• Download powerpoint

FIG. 5.

The CK2 site mutant Z(S167A/S173A) is reduced in its capacity to bind a ZEBRA response element in vitro. (A) Immunoblot probed for ZEBRA present in 0.42 M NaCl cell extracts used for the EMSA experiments illustrated in panel B. (B) EMSA using serial dilutions of BZKO cell extracts transfected with the wild type or the Z(S167A/S173A) mutant. (C) Immunoblot of titrated cell extract expressing the Z(S167A/S173A) mutant used in the experiments illustrated in panel D. (D) EMSA comparing DNA binding by wt ZEBRA and the Z(S167A/S173A) mutant, using equal amounts of immunoreactive ZEBRA. NS, nonspecific.

The CK2 site mutant Z(S167A/S173A) is defective in association with EBV DNA in vivo.To compare the associations of wt ZEBRA and the Z(S167A/S173A) mutant with EBV DNA in vivo, we carried out ChIP experiments (Fig. 6 and 7). In preliminary experiments, end-point PCRs (28 cycles) were performed to estimate the amount of oriLyt DNA that was immunoprecipitated by antibody to ZEBRA (Fig. 6A). Less oriLyt was immunoprecipitated from BZKO cells transfected with the CK2 mutant than from cells transfected with the wt ZEBRA protein (Fig. 6A, compare lanes 8 and 9). Equal amounts of immunoreactive ZEBRA protein could be immunoprecipitated under conditions that were identical to those used for ChIP (Fig. 6B). Under these PCR conditions, there was slightly less input EBV DNA amplified from BZKO cells that were transfected with the CK2 mutant (Fig. 6A, compare lanes 2 and 3). This result was anticipated by the experiment illustrated in Fig. 2 showing that the CK2 mutant was deficient in amplifying EBV DNA. To compare quantitatively the amounts of ZEBRA protein that associated with EBV DNA, it was essential to limit EBV DNA amplification following introduction of the wild-type protein. This was accomplished in subsequent experiments by carrying out all transfections in the presence of phosphonoacetic acid (PAA) (Fig. 7).

• Open in new tab
• Download powerpoint

FIG. 6.

Reduced association of the Z(S167A/S173A) mutant with oriLyt in vivo. (A) End-point PCR (28 cycles) of ChIP products, assaying for binding of the wild type or the Z(S167A/S173A) mutant to oriLyt. (B) Immunoprecipitation of wt ZEBRA and the Z(S167A/S173A) mutant under ChIP conditions. The immunoprecipitate was analyzed by use of a polyclonal anti-ZEBRA antibody on an immunoblot.

• Open in new tab
• Download powerpoint

FIG. 7.

Quantitation of the relative levels of association of wt ZEBRA and the Z(S167A/S173A) mutant with EBV DNA, as measured by ChIP. (A) BZKO cells were transfected with the indicated expression plasmids in the presence of PAA to inhibit viral DNA amplification. Immunoprecipitation was carried out with preimmune serum or a polyclonal antibody to ZEBRA. The amount of viral DNA precipitated was quantitated by real-time PCR with primers that detect EBV oriLyt. (B) Summary of three ChIP experiments comparing the associations of ZEBRA and the Z(S167A/S173A) mutant with EBV oriLyt DNA in the presence of PAA. (C) Chromatin immunoprecipitation experiment comparing the associations of wt ZEBRA, the Z(S186E) mutant, and the Z(S167A/S173A) mutant with Rp, the promoter of the EBV BRLF1 gene.

Under these conditions, the amounts of input viral DNA used for ChIP were similar for cells transfected with empty vector (cytomegalovirus), the DNA binding-deficient Z(S186E) mutant, wt ZEBRA, and the Z(S167A/S173A) mutant (Fig. 7A). Antibody to ZEBRA immunoprecipitated 2.9-fold less oriLyt DNA in cells transfected with the Z(S167A/S173A) mutant than in cells transfected with wt ZEBRA and 9.9-fold less oriLyt DNA in cells transfected with the DNA binding-deficient Z(S186E) mutant than in cells transfected with wt ZEBRA. In three replicate experiments, the wt ZEBRA protein associated with oriLyt DNA approximately 3-fold more efficiently than the CK2 site mutant Z(S167A/S173A) (range, 2.4- to 4.5-fold), with correction for the amount of input DNA. The deficiency in association of the Z(S167A/S173A) mutant with EBV DNA was of a similar magnitude whether primers for oriLyt (Fig. 7B) or Rp, the promoter of the BRLF1 gene, were employed (Fig. 7C).

The Z(S167A/S173A) mutant is more soluble than wt ZEBRA protein.The total amount of immunoreactive ZEBRA protein expressed in BZKO cells was the same whether a wild-type or a Z(S167A/S173A) mutant expression plasmid was transfected (Fig. 8A). However, the mutant was dramatically more soluble than the wild-type protein in cell extracts prepared in 0.3 M or 0.42 M NaCl and slightly more soluble in 0.6 M NaCl (Fig. 8B). The difference in solubility between wt ZEBRA and the Z(S167A/S173A) mutant was maintained over a period of 72 h after transfection of BZKO cells with expression plasmids (Fig. 8C). This difference was specific to ZEBRA, since the BMFR1 EA-D protein was equally soluble in 0.3 M NaCl extracts of cells that expressed wild-type or S167A/S173A mutant ZEBRA protein (Fig. 8C).

• Open in new tab
• Download powerpoint

FIG. 8.

The CK2 site mutant Z(S167A/S173A) is more soluble than wt ZEBRA in BZKO cell extracts prepared in 0.3 M to 0.6 M NaCl. (A) Total cell extracts prepared in SDS sample buffer. (B) Effects of different NaCl concentrations on the solubility of wt ZEBRA and the Z(S167A/S173A) mutant. For panels A and B, the levels of ZEBRA protein were determined by immunoblotting with a polyclonal antibody to ZEBRA. (C) BZKO cells were transfected with an expression vector for ZEBRA or the Z(S167A/S173A) mutant. At daily intervals after transfection, cell extracts were prepared in 0.3 M NaCl and analyzed for the levels of ZEBRA and EA-D by immunoblotting.

DNase I treatment releases ZEBRA from the insoluble fraction.The result from the previous experiment showing that wt ZEBRA was less soluble than the Z(S167A/S173A) mutant in 0.3 M NaCl suggested that the wt ZEBRA protein has a higher affinity than the mutant for an insoluble macromolecular complex. To test whether DNA binding activity was required to maintain the association of the ZEBRA protein with the insoluble fraction, we examined the solubilities of six DNA binding-deficient mutants [Z(K178E), Z(R179E), Z(K181E), Z(N182E), Z(R183E), and Z(S186E)] (18). All six point mutations of ZEBRA were located in the DNA binding domain. A strict correlation was apparent between the DNA binding activity of ZEBRA and its solubility in 0.3 M salt (data not shown). DNA binding-deficient mutants were invariably more soluble than wt ZEBRA. To determine whether DNA was an essential component of the complex that retains ZEBRA in the insoluble fraction, ZEBRA-expressing cell extracts were treated with DNase I for different times (30, 60, and 120 min). The addition of DNase I led to a 47-fold increase in soluble wt ZEBRA protein; DNase I treatment increased the solubility of the Z(S167A/S173A) mutant only 1.9-fold, since a large fraction of the mutant protein was already soluble in the absence of DNase I (Fig. 9). This result indicated that DNA is an essential component of the macromolecular complex that renders ZEBRA resistant to extraction with 0.3 M NaCl. The experiment corroborates, using an alternative assay, the finding by ChIP (Fig. 7) that the Z(S167A/S173A) mutant is deficient in association with DNA in vivo.

• Open in new tab
• Download powerpoint

FIG. 9.

Solubility of wt ZEBRA is enhanced by treating cell extracts with DNase I. BZKO cells were transfected with wt ZEBRA or the Z(S167A/S173A) mutant. Twenty-four hours after transfection, cells were resuspended in 0.3 M NaCl EMSA buffer and then diluted 1:5 in EMSA buffer without NaCl. Extracts without (−) or with (+) DNase I were incubated for the indicated times. An immunoblot was probed with a polyclonal antibody to ZEBRA. CMV, cytomegalovirus.

A phosphomimetic negative charge at position 173 preserves the viral DNA replication function of ZEBRA.To explore further the importance of phosphorylation at position 173 in the capacity of the ZEBRA protein to initiate lytic viral DNA replication, serine 173 was mutated to aspartic acid. The ZEBRA mutant with this phosphomimetic substitution activated expression of the late protein FR3 (Fig. 10A, lane 5) and maintained the ability of ZEBRA to activate lytic viral DNA replication (Fig. 10B, lanes 5 and 10). ZEBRA containing an alanine substitution at position 173 reproducibly activated early gene expression of EA-D but failed to activate viral replication and late gene expression of BFRF3 (Fig. 10A and B, lanes 3, 4, 8, and 9). The efficiency of the Z(S173D) mutant in the activation of viral genome amplification was quantitated using real-time PCR with primers directed towards the origin of lytic replication. Wild-type ZEBRA amplified the viral genome 45-fold relative to empty vector, and the Z(S173D) mutant amplified the viral genome 71-fold.

• Open in new tab
• Download powerpoint

FIG. 10.

Phosphomimetic substitution of ZEBRA (S173) maintains the capacity of ZEBRA to activate lytic DNA replication and viral late gene expression. BZKO cells were transfected with the indicated expression plasmids for wild-type or mutated ZEBRA proteins. (A) Immunoblots were probed with antibodies to the indicated EBV proteins. (B) A Southern blot prepared from total cellular DNA harvested 48 h and 72 h after transfection was probed with the Xho 1.9 fragment (34). (C) Quantitative real-time PCR measuring the amounts of viral DNA amplification 48 h after transfection of the indicated ZEBRA proteins.

Aspartate substitution at position 173 results in a solubility and DNA binding phenotype analogous to that of wt ZEBRA.To investigate the effects of an aspartate mutation at amino acid 173 on the solubility characteristics of ZEBRA, wt ZEBRA and the Z(S173A), Z(S173D), and Z(S167A/S173A) mutants were expressed in BZKO cells, and their solubilities were compared (Fig. 11). In total cell SDS extracts, all three ZEBRA mutants were expressed to the same level as the wild-type protein (Fig. 11A). Wild-type ZEBRA and the phosphomimetic mutant, Z(S173D), were markedly less soluble in 0.3 M NaCl than the Z(S173A) mutant (Fig. 11B). Treatment of cell extracts with DNase I for 30 min resulted in 47-, 1.4-, and 18.2-fold increases in the solubilities of wt ZEBRA and the Z(S173A) and Z(S173D) mutants, respectively.

• Open in new tab
• Download powerpoint

FIG. 11.

The phosphomimetic substitution mutant Z(S173D) maintains the solubility and DNA binding characteristics of wt ZEBRA. (A) Western blot analysis of total cell extracts expressing wild-type or mutant ZEBRA, using an antibody against ZEBRA. (B) DNase I treatment increases the solubility of the Z(S173D) mutant. (C) Comparison of the solubilities of wt ZEBRA and CK2 site mutants at two concentrations of NaCl, 0.3 M and 1 M. (D) Comparison of the capacity of wt ZEBRA and the CK2 site mutants of ZEBRA extracted in 0.3 M and 1 M NaCl to bind to ZRE-2 in an EMSA experiment.

The main hurdle in comparing the in vitro DNA binding activities of the Z(S173D), Z(S173A), and Z(S167A/S173A) mutants with that of wt ZEBRA was the presence of different levels of ZEBRA protein solubilized in 0.42 M NaCl, the salt concentration commonly used to prepare EMSA extracts. To overcome this problem, ZEBRA and the CK2 site mutants were extracted with two salt concentrations (0.3 and 1 M NaCl), and the relationship between the amount of ZEBRA extracted and the DNA binding activity was examined (Fig. 11C and D). At 0.3 M NaCl, wt ZEBRA and the Z(S173D) mutant were less soluble than the Z(S173A) and Z(S167A/S173A) mutants. However, wt ZEBRA and the Z(S173D) mutant prepared in 0.3 M NaCl bound the ZRE-2 site slightly better than the other two mutants. Extraction of ZEBRA with EMSA buffer containing 1 M NaCl increased the solubilities of wt ZEBRA and the Z(S173D) mutant and yielded ZEBRA proteins at relatively equal levels. Under these conditions, wt ZEBRA and the Z(S173D) mutant bound a ZRE-2 site with at least twice the affinity of the Z(S173A) and Z(S167A/S173A) mutants (Fig. 11C and D).

Phenotype of the Z(S173A) mutant.The EBV lytic cycle can be divided into the following five main phases: induction of ZEBRA and Rta expression, downstream early gene expression, DNA replication, late gene expression, and virion assembly and egress. ZEBRA plays a major role in activating Rta, stimulating early gene expression and initiating viral replication. Transactivation of the BRLF1 gene (Rta) is solely dependent on the action of ZEBRA. However, the expression of two downstream genes, the BMRF1 and BHRF1 genes, requires the synergistic action of both ZEBRA and Rta. Thus, ZEBRA uses at least two modes of transcription activation: one is independent of Rta, and the other involves synergy with Rta. Our data demonstrate that mutation of the major CK2 substrate site, S173, to alanine or aspartate does not enhance or diminish the ability of ZEBRA to activate the expression of Rta (BRLF1) or EA-D (BMRF1) from the endogenous viral genome. This result suggests that phosphorylation of ZEBRA at S173 is not required either for the autonomous or for the synergistic mode by which ZEBRA activates the transcription of early lytic cycle genes. However, it was conceivable that ZEBRA activates another set of viral genes via a third mode of transcriptional activation that is contingent on ZEBRA phosphorylation at S173. EBV encodes many components of the lytic replication machinery. Expression of ZEBRA activates, directly or indirectly, the transcription of these replication genes. Therefore, we compared the mRNA levels of different components of the viral replication machinery (BALF2, BALF5, BBLF2/3, BBLF4, and BSLF1) in cells transfected with wt ZEBRA, the Z(S173A) mutant, or the Z(S173D) mutant. We found that the Z(S173A) mutant was competent to stimulate the expression of each of these genes (Fig. 3). Therefore, the defect in replication observed with the Z(S173A) mutant is unlikely to be due to a lack of expression of one of these essential virally encoded replication proteins. Nevertheless, the possibility remains that the Z(S173A) mutant has a defect in activating the transcription of other viral or cellular genes that we did not study. These genes might play a significant role in EBV lytic replication.

Pros and cons regarding the role of phosphorylation at serine 173 in enhancing the DNA binding activity of ZEBRA.A number of observations favor the role of phosphorylation at S173 in enhancing the DNA binding activity of ZEBRA. These observations are as follows. (i) S173 is a target for phosphorylation in vitro by CK2 (7, 24). (ii) S173 is phosphorylated in vivo by an unknown kinase, which is likely CK2 (7). (iii) A correlation exists between the extent of phosphorylation by CK2 in vitro and the phenotypes of mutants with substitutions in the two CK2 substrate sites in ZEBRA, S167 and S173. S167 is weakly phosphorylated by CK2, and the S167A mutant has nearly wild type phosphorylation (Fig. 2). S173 is the major phosphoacceptor site for in vitro phosphorylation by CK2, and the S173A mutant is deficient in DNA binding and replication. (iv) The S173A mutant, which abolishes phosphorylation in vitro and in vivo, is impaired in DNA binding in vitro (by EMSA) and in vivo (by ChIP) and is replication deficient. (v) S173 is located upstream of the known DNA binding domain and is not likely to contact DNA directly (32). Therefore, S173 is presumably regulatory in its behavior. (vi) The S173D mutant, a phosphomimetic mutant, is not impaired in DNA binding (EMSA) and is replication competent (Fig. 10 and 11). (vii) However, any aspartate substitution in the DNA binding domain itself invariably inhibits DNA binding (18). (viii) Tetrabromobenzimidazole, a specific inhibitor of CK2, blocks EBV lytic replication (data not shown).

There are three potential arguments against the essential role of phosphorylation of S173 in enhancing DNA binding. All of these arguments are derived from studies of a ZEBRA protein that was expressed in Escherichia coli in a presumably unphosphorylated state (9, 14, 24), and they are as follows. (i) The bacterially expressed ZEBRA protein maintains its capacity to bind DNA in vitro. (ii) The S173A mutation in ZEBRA expressed in E. coli inhibits DNA binding. (iii) When ZEBRA expressed in E. coli was phosphorylated in vitro with CK2 purified from bovine liver, its capacity to bind DNA was inhibited.

The differences observed in DNA binding activities of ZEBRA proteins synthesized in prokaryotic and eukaryotic expression systems could be attributed to posttranslational modifications. In a bacterial expression system, ZEBRA is presumably unmodified and, in particular, unphosphorylated. In its unmodified form, DNA binding would be the default mode. In a eukaryotic expression system, posttranslational modifications of ZEBRA might have both positive and negative effects on DNA binding. Phosphorylation of S173 might be crucial to overcome the negative effects of other posttranslational modifications. DNA binding might be influenced by multiple effects exerted by different posttranslational modifications. Support for this model comes from a previous study with bacterial protein where simultaneous phosphorylation of S167 and S173 inhibited the capacity of ZEBRA to bind DNA (24). We have also found in eukaryotic cells that phosphomimetic mutants with substitutions at both S167 and S173 exhibited the same phenotype as the S173A mutant, namely, the failure to activate viral replication (data not shown). Both of these experiments suggest that phosphorylation of S167 has a dominant inhibitory effect on DNA binding. Since S167 and S173 are located upstream of the DNA binding domain, it is likely that posttranslational modifications of these two residues interact to affect the conformation of ZEBRA to promote or inhibit DNA binding. The interaction between the two phosphoacceptor CK2 sites, S167 and S173, can be studied further using monoclonal antibodies directed against the phosphorylated form of each site.

Possible role of phosphorylation in regulating origin recognition by ZEBRA.Previous studies of prokaryotic and eukaryotic OBPs revealed that noncovalent association with ATP modifies the DNA binding activity of this group of proteins. In S. cerevisiae, the ORC specifically interacts with two elements in the yeast origin of replication, the A and B1 elements (29, 36). Such protein-origin interactions are dependent on the presence of an intact ATP binding site in the ORC1 protein (22). Similarly, in prokaryotes, only ATP-bound DnaA is capable of initiating bacterial replication (40). ATP-DnaA specifically binds to five binding sites located in the bacterial origin of replication, oriC. Initial binding of ATP-DnaA to oriC triggers a cooperative association of DnaA monomers with the AT-rich region of oriC, thus facilitating origin unwinding (44, 45).

Unlike ORC and DnaA, ZEBRA lacks canonical sequences for an ATP binding domain (GXXXXGKS/T). However, in a process analogous to origin recognition by ATP-bound ORC and DnaA, only the S173-phosphorylated form of ZEBRA can tightly bind to oriLyt and initiate lytic viral replication. A lack of phosphorylation at this residue reduced the overall DNA binding capacity of ZEBRA, as assayed in vitro by EMSA and in vivo by chromatin immunoprecipitation. Moreover, a phosphomimetic mutant with a substitution at position 173 exhibited DNA binding activity similar to that of wt ZEBRA. Consistent with this functional similarity between ATP binding and phosphorylation at S173, mutational analysis of the ATP binding domain of DnaA identified two amino acids, K178 and N235, as crucial residues for the protein's affinity for ATP or ADP. Replacement of these two residues with isoleucine and asparagine, respectively, reduced bacterial replication 10-fold and inhibited opening of the DNA duplex at oriC (30). Similarly, in S. cerevisiae, mutations in the Walker A motif of ORCp1, the ATP binding motif, disrupted ORC binding to the origin of DNA replication (28).

As an origin binding protein, ZEBRA binds to seven ZEBRA response elements (ZREs 1 to 7) in oriLyt. ZREs 1 to 4 are located in the upstream component of oriLyt, while ZREs 5 to 7 are present in the downstream region of oriLyt. Mutational analysis of oriLyt revealed that ZREs 1 to 4, but not ZREs 5 to 7, are indispensable for activation of viral replication (38). In our studies, assessment of the amount of ZEBRA bound to the upstream region of oriLyt using chromatin immunoprecipitation revealed that the Z(S173A) mutant consistently exhibited a threefold defect in the capacity to associate with oriLyt in vivo compared to wt ZEBRA. This defect could be attributed to a generally weak DNA binding activity of the Z(S173A) mutant or to an inability of the Z(S173A) mutant to interact with a specific ZEBRA binding site in oriLyt. Thus, the mutant protein may either randomly occupy one or two sites on different origins of lytic replication or weakly bind to an entry site and fail to recruit other ZEBRA homodimers that cooperatively occupy oriLyt (Fig. 12A). Alternatively, the Z(S173A) mutant might bind and recruit other ZEBRA molecules to oriLyt, but only for a small subpopulation of EBV molecules as a result of its weak affinity for DNA (Fig. 12B). The affinity of ZEBRA for DNA may also be influenced by protein-protein interactions. Further work is needed to investigate the role of phosphorylation at S173 in interactions of ZEBRA with other replication proteins and transcription factors.

• Open in new tab
• Download powerpoint

FIG. 12.

Proposed models portraying the mechanism by which phosphorylation of ZEBRA at S173 might regulate the occupancy of ZEBRA binding sites in oriLyt. In model A, if ZREs are bound independently, then the Z(S173A) mutant would not saturate all ZREs on any EBV molecule. In model B, if ZREs are bound cooperatively, then the Z(S173A) mutant would occupy only a subpopulation of molecules. A circled “P” depicts a form of ZEBRA phosphorylated at S173. A boxed “OH” depicts a form of ZEBRA that is not phosphorylated at S173. Numbers 1, 2, and 3 represent three different EBV genomes.

Solubility as a measure of ZEBRA's affinity for DNA in vivo.We found that the DNA binding activity of ZEBRA was inversely proportional to the solubility of the protein in cell suspensions prepared using EMSA lysis buffer containing 0.3 M NaCl. DNA binding-competent ZEBRA proteins, such as wt ZEBRA, the Z(S167A) mutant, and the phosphomimetic mutant, Z(S173D), were mostly insoluble in 0.3 M NaCl. DNA binding-defective ZEBRA mutants, such as the Z(S173A) and Z(S167A/S173A) mutants, were significantly more soluble at the same salt concentration. This correlation between DNA binding and solubility was further corroborated by examining the solubilities of six non-DNA-binding single point mutations, all of which resided in the DNA binding domain of ZEBRA. All six mutants exhibited markedly higher solubilities in 0.3 M NaCl extraction buffer than did wt ZEBRA (A. El-Guindy et al., unpublished data). To demonstrate that DNA is an essential component of the macromolecular complex that retains ZEBRA in the insoluble fraction, we compared the solubilities of ZEBRA before and after treatment with DNase I. We found that the wild type and the Z(S167A), Z(S173A), Z(S167A/S173A), and Z(S173D) mutants became equally soluble following DNase I treatment of cell extracts prepared in 0.3 M NaCl. This correlation between the DNA binding activity of ZEBRA and the solubility of the protein in 0.3 M NaCl provided an independent means to assess the affinity of ZEBRA for DNA in vivo. Thus, the observed increase in the solubility of the S173A mutant compared to both wt ZEBRA and the phosphomimetic mutant provides additional evidence that the Z(S173A) mutant exhibits a lower affinity for DNA.

Replication versus transcription.Activation of lytic viral replication by ZEBRA demands a highly specific DNA binding activity that allows ZEBRA to locate oriLyt among cellular and viral DNAs. Likewise, the ability of ZEBRA to activate transcription is dependent on its capacity to bind elements in ZEBRA-responsive promoters. This raises an important question, namely, how can ZEBRA distinguish between transcription activation sites located in target promoters and replication initiation sites located in oriLyt? Here we found that a reduction in the DNA binding activity of the Z(S173A) mutant had no significant effect on the capacity of ZEBRA to activate transcription. A similar, threefold defect relative to wt ZEBRA was observed when Rp cross-linked to the Z(S173A) mutant was quantitatively analyzed by real-time PCR (Fig. 7C). However, an equivalent reduction in the association of the Z(S173A) mutant with oriLyt markedly abridged ZEBRA's role in the activation of viral replication. This obligatory requirement for strong DNA binding activity by ZEBRA during initiation of viral replication, not transcription activation, could be explained by the ability of ZEBRA to occupy all four sites in oriLyt. If the relative binding capacity of the Z(S173A) mutant is 0.3 compared to wt ZEBRA, then the probability that the mutant will simultaneously occupy all four sites on a single molecule is 0.0081. However, if only one binding site is required for transcriptional activation, then the probability of the Z(S173A) mutant binding this site is 0.3, which may be adequate to activate transcription. Alternatively, this difference between transcription and replication could also be influenced by protein-protein interactions, localization to replication compartments, or topological changes induced by ZEBRA binding to oriLyt, such as origin bending. In adenovirus type 5 (Ad5), two cellular transcription factors, namely, nuclear factor I and octamer binding protein (Oct-1), bind in close proximity to their recognition sequences in the Ad5 origin of replication and induce an 82° collective bend (31). Further work is needed to investigate the role of phosphorylation at S173 in the interactions of ZEBRA with oriLyt and other replication proteins.

Phosphorylation of ZEBRA and the EBV lytic cycle.Previously, we demonstrated that the ability of ZEBRA to repress Rta-mediated activation of the late LR2 protein was dependent on ZEBRA's phosphorylation at position 173 (7). Based on these findings, we proposed a role for phosphorylation of ZEBRA in the temporal regulation of the EBV lytic cycle. We suggested that phosphorylation of ZEBRA at this residue persists during the early phase of the lytic cycle. Here we provide additional evidence in support of our previous hypothesis. This evidence suggests that during the early phase, phosphorylation of ZEBRA at S173 is not only required to maintain the proper temporal order of lytic gene expression but is also essential for initiation of viral replication.